Length dependence of charge transport in nanoscopic molecular junctions incorporating a series of rigid thiol-terminated norbornylogs

Article abstract of DOI:10.1021/jp044630p

Four tetrathiol-terminated norbornane homologues were synthesized and self-assembled monolayers (SAMs) of these molecules were formed on Au via adsorption from CH₂Cl₂. SAMs were characterized structurally via spectroscopic ellipsomet

Full text of DOI:10.1021/jp044630p

J. Phys. Chem. B 2005, 109, 5207-5215
5207
Length Dependence of Charge Transport in Nanoscopic Molecular Junctions Incorporating
a Series of Rigid Thiol-Terminated Norbornylogs
†
‡
§
§
§
Jeremy M. Beebe, Vincent B. Engelkes, Jingquan Liu, J. Justin Gooding, Paul K. Eggers,
†
†
,§
,‡
Yongseok Jun, Xiaoyang Zhu, Michael N. Paddon-Row,* and C. Daniel Frisbie*
Department of Chemistry, UniVersity of Minnesota, 209 Pleasant Street SE, Minneapolis, Minnesota 55455,
Department of Chemical Engineering and Materials Science, UniVersity of Minnesota, 421 Washington
AVenue SE, Minneapolis, Minnesota 55455, and School of Chemistry, UniVersity of New South Wales,
Sydney, NSW 2052, Australia
ReceiVed: NoVember 24, 2004; In Final Form: January 13, 2005
Four tetrathiol-terminated norbornane homologues were synthesized and self-assembled monolayers (SAMs)
of these molecules were formed on Au via adsorption from CH Cl . SAMs were characterized structurally
2
2
via spectroscopic ellipsometry (SE), reflection-absorption infrared spectroscopy (RAIRS), Rutherford
backscattering spectrometry (RBS), and X-ray photoelectron spectroscopy (XPS). Results of these analyses
show that the rigid norbornylogs form monolayers that have a surface coverage slightly lower than that of
alkanethiols, and that they exhibit a nonmonotonic dependence of film thickness on molecular length. Nanoscale
molecular junctions incorporating these SAMs were formed and characterized electrically using conducting
probe atomic force microscopy (CP-AFM). The resistances of these junctions scale exponentially with the
-
1
contour length of the molecules, with â ) 0.9 Å , consistent with a nonresonant tunneling mechanism.
Further, the resistance of norbornyl SAMs correlates well with the resistance of alkanedithiol SAMs of similar
length, suggesting that the norbornyl molecules form sulfur-metal bonds on both ends of the junction.
Introduction
in the cases we have examined). We have also shown that the
resistance versus length characteristic can be used to provide
estimates of contact resistance.
The formation and characterization of metal-molecule-metal
junctions is of central importance to molecular electronics.
Several strategies exist for probing the electrical characteristics
of molecules by sandwiching them between metal electrodes.
2
9,30
Here we report CP-AFM tunneling measurements through
self-assembled monolayers of a new series of norbornylogous
molecules 1-4, shown in Figure 1. These molecules are long,
rigid, and curved. In principle, these characteristics make these
molecules good candidates for determining whether transport
occurs via a through-bond mechanism, in which charge travels
along the contour length of the molecule, or a through-space
mechanism, in which charge tunnels in a straight-line path from
one electrode to the other. The debate over through-bond or
through-space tunneling has occurred repeatedly in chemistry,
1
,2
3,4
These methods include break junctions, nanopores, crossed
wires,5 Hg-drop electrodes,
,6
7-10
scanning tunneling microscopy
1
1-15
(
(
STM),
CP-AFM).
and conducting probe-atomic force microscopy
1
6-20
Already a wide range of electrical transport
behaviors have been observed, including negative differential
resistance,1
4,21,22
rectification,
6,7,23
conductance quantization,
In general, the goal of making conductance
24,25
9,26,27
and switching.
measurements on molecules in nanoscale junctions is to examine
the effects of structure and electronic properties on the charge
transport characteristics of the molecular bridge.
Recently, we have used CP-AFM to make molecular junctions
based on self-assembled monolayers of alkanethiols, alkane-
3
1-33
for example in the discussion of electron transfer in proteins,
3
4-36
37-42
DNA,
and donor-bridge-acceptor compounds.
The
preponderance of evidence in these systems points to through-
bond tunneling as the dominant mechanism. Prior work by one
of us involving molecules similar to the norbornylogs used in
2
8
dithiols,29 alkyl isocyanides, and oligophenylenethiols. In this
technique, SAMs are grown on a metal substrate that is brought
into contact with a metal-coated AFM tip under controlled load.
Typically, a bias is swept at the tip, and the substrate is held at
ground, allowing the current-voltage characteristics of the
junction to be measured. At low bias, the current-voltage curves
are linear, and the reciprocal of the slope is the low-bias junction
resistance. We use CP-AFM in this manner to measure junction
resistance as a function of molecular length. The resistance-
length characteristic quantifies the efficiency of transport and
reflects the transport mechanism (usually nonresonant tunneling
30
18
this study overwhelmingly supports a through-bond tunneling
37-42
mechanism in solution.
However, rigid norbornylogs have
not yet been exploited as functional components in molecular
junctions.
One factor that affects electron transport in the solid state is
the arrangement of molecules on the substrate surface. As
discussed later, SAMs of the norbornylogs display a nonmono-
tonic dependence of SAM thickness on the number of carbon
atoms in the molecular backbone. At first glance, this should
provide a pathway for distinguishing between through-bond and
through-space charge transport in molecular junctions, because
it should be possible to examine whether conductance scales
with the number of carbon atoms or with SAM thickness.
However, after extensive monolayer characterization, we show
that a change in molecular orientation occurs within the SAMs,
†
Department of Chemistry, University of Minnesota.
Department of Chemical Engineering and Materials Science, University
‡
of Minnesota.
§
School of Chemistry, University of New South Wales.
1
0.1021/jp044630p CCC: $30.25 © 2005 American Chemical Society
Published on Web 03/01/2005

5208 J. Phys. Chem. B, Vol. 109, No. 11, 2005
Beebe et al.
polarization angles (Ψ and ∆) were measured for both bare Au
substrates and SAMs over the wavelength range 250-900 nm
at an incident angle of 75° from the surface normal. Prior to
SAM deposition, the optical constants n and k (the index of
refraction and extinction coefficient, respectively) were calcu-
lated for each individual Au substrate using instrument software
to convert the Ψ and ∆ data assuming a bulk solid (i.e., infinite
thickness). After SAM deposition, the polarization measure-
ments were repeated with each sample in the same orientation
as when the bare Au substrate measurement was taken. Using
the new Ψ and ∆ values and fixed values for n and k (1.5 and
0
, respectively), the instrument software calculated the film
thickness. The calculation is based on a parallel layer model of
substrate/film/ambient that assumes the film is homogeneous
and isotropic. Reported thicknesses represent an average over
approximately 10 different film formations.
SAM Thickness and Orientation by XPS. X-ray photo-
electron spectra were taken on a Perkin-Elmer Phi 5400
spectrometer with the Mg KR X-ray (1253.6 eV) from an anode
source using a hemispherical analyzer in an ultrahigh vacuum
system. Each sample was transferred from the ambient into the
UHV chamber through a high vacuum cell. The X-ray anode
was operated at 200 W, and the analyzer at a pass energy of
89.45 eV for survey scans and 17.9 eV for high-resolution scans.
Thickness calculations from XPS data for thin films can be
calculated on the basis of the attenuation of the Au 4f signal
from the substrate. This attenuation is given by
Figure 1. Structures of the norbornylogous molecules 1-4. Due to
the synthetic procedures used, each tetrathiol system is, in fact, an
equimolar mixture of two diastereomers; see the Supporting Informa-
tion.
thus complicating the assignment of a transport mechanism.
Nevertheless, the monolayer characterization studies indicate
that these molecules form reasonably high-quality monolayers.
Their ladder-like structure makes them candidates for a wide
range of molecular electronics experiments where rigid mol-
ecules are desired.
I
Z
)
exp -
(1)
(
)
I0
λ cos θ
Experimental Section
where Z is the film thickness, I and I0 an average of the
intensities of the Au 4f5/2 and 4f7/2 peaks after and before
monolayer growth, respectively; θ is the angle of photoelectron
detection (from surface normal); λ is the effective attenuation
length of the photoelectron through the carbon chain. In these
experiments, θ was 45° and the effective attenuation length of
Fabrication of Au Substrates. Au substrates were fabricated
in a manner similar to earlier experiments.1
8,28-30,43
Si (100)
wafers were coated with 50 Å Cr and 1000 Å Au in a thermal
-
6
evaporator at a pressure of 1 × 10 Torr. The resulting Au
surfaces exhibited an RMS roughness of ∼2 nm over an area
2
of 1 µm .
49,50
photoelectrons was taken to be 34.63 Å.
Sample Preparation and Storage. Synthetic schemes for
the formation of the norbornylogous molecules 1-4 (Figure 1)
can be found in the Supporting Information. Monolayer films
were initially formed in the same manner as alkanethiolssAu
substrates were exposed to millimolar solutions of the molecules
in CH2Cl2 for 12 h. Thicknesses of films deposited in this
manner ranged from 20 to 75 Å for 1, far too thick to be a
single molecular layer. The same was true for the other chain
lengthssall molecules preferentially assemble into terlayers or
tetralayers if left in solution for 12 h, likely due to disulfide
In addition to measuring film thicknesses, XPS analysis was
also used to determine the binding energies of various sulfur
species. In these determinations, high-resolution scans were
performed on the S 2p region. Because the binding energies
associated with free thiol and bound thiol are so similar, a peak
fitting routine was used to deconvolute the two signals. Fits to
the data used a combination of Gaussian and Lorentzian line
shapes. To fit the S 2p3/2,1/2 doublet, we used a pair of peaks
with the same full width at half-maximum, the standard spin-
orbit splitting of 1.17 eV, and a branching ratio of 0.5555
(S 2p /S 2p ). An oxidized sulfur species with a binding
4
4-48
formation.
This multilayering was controlled by moving
1
/2
3/2
to a short deposition period followed by a longer soak period.
Films were also deposited from less concentrated solutions. Once
the deposition parameters were modified in this manner, the
molecules reproducibly assembled into single monolayers on
Au. The final deposition conditions for SAM formation were
that Au substrates were exposed to 10 µM solutions of molecules
energy of 167.9 eV contributes approximately 10% of the total
peak area. We have assumed that the oxidized sulfur peak arises
solely from sulfur atoms that are not bound to the metal surface.
The XPS system used for analysis is not equipped with an X-ray
monochromator, which makes complete peak separation impos-
sible, due to the decreased energy resolution. These two factors
combine to make the error in the measurement of peak areas at
least 20%.
1
-4 for 1 h, after which they were soaked in neat CH2Cl2 for
2
4 h. Samples were stored in CH2Cl2 until analysis by
spectroscopic ellipsometry, reflection-absorption infrared spec-
Infrared Spectroscopy. KBr pellet and RAIRs measurements
were performed on a Nicolet Series II Magna-IR System 750
FTIR. RAIRS experiments on SAMs of 1-4 used a Harrick-
Seagull accessory with the IR beam incident at an 86° angle
with respect to the surface normal. A total of 2000 scans were
troscopy (RAIRS), Rutherford backscattering spectrometry
(RBS), X-ray photoelectron spectroscopy (XPS), or conducting
probe-AFM. Upon removal from CH2Cl2, the samples were
rinsed in neat ethanol or CH2Cl2 and blown dry with N2.
SAM Thickness by Spectroscopic Ellipsometry. Ellipso-
metric measurements were made using a Sopra ES4G spectro-
scopic ellipsometer equipped with a xenon light source. The
-1
collected at 4 cm resolution. KBr pellet spectra were measured
-1
at 0.5 cm resolution with 100 scans. Molecular concentration
-
3
-3
in the pellets ranged between 3.0 × 10 and 9.0 × 10 mol

SAMs of Norbornane Homologues on Au
J. Phys. Chem. B, Vol. 109, No. 11, 2005 5209
-
3
cm . Peak areas were determined by baseline correcting the
spectra and fitting the peaks with Lorentzian line shapes.
+
Rutherford Backscattering Spectrometry. He ions for
RBS measurements were generated using a MAS 1700 pelletron
tandem ion accelerator (5SDH) equipped with a charge exchange
RF plasma source by National Electrostatics Corporation (NEC).
The software (HYPRA) and analysis equipment (RBS 400) were
provided by Charles Evans & Associates. Ions were accelerated
at 2 MeV with a beam current of 10 nA. The solid angle
subtended by the backscattered ion detector was 3.6 msr. The
detector was placed at a 165° angle with respect to the incident
beam on the sample. Silicon (100) substrates for monolayer
formation were coated with 400 Å Au without a Cr adhesion
layer for an improved RBS signal-to-background ratio. Polar
scans were first performed in the silicon region to find the
sample orientation that maximized channeling effects. This
minimized the background signal from the single-crystal Si
substrate (and thus maximized the overall signal-to-noise ratio
in the non-Si regions of the spectrum). Charge (200 µC) was
then collected in increments of 20 µC at one point on each
sample; the total collection time was ∼5 h per sample. Data
were taken in increments in case film degradation would occur
during an extended measurement. Degradation was not observed
as determined by a constant sulfur signal across the 10
acquisitions. Molecular surface coverage was determined by
dividing the measured sulfur coverage by four, as there are four
sulfur atoms per molecule.
Figure 2. B3LYP/6-31G(d) minimization of structures related to the
norbornylogous molecules used in this study. CH OCH has been
2 3
replaced with a methyl group for computational expediency. Also shown
is the calculated molecular contour length.
Figure 3. Average of 10-20 thickness measurements obtained by
ellipsometry (filled circles) increasing monotonically with molecular
contour length for SAMs of 1-3. Extrapolation of this trend shows an
Electrical Characterization by CP-AFM. Two distinct types
of experiments were performed in this study. One set of
experiments was designed to compare transport in the norbornyl
homologues with transport in alkanethiols and -dithiols. The
AFM tips used for this set of experiments had a nominal radius
of 20 nm, and were purchased from Digital Instruments. The
coating procedure was the same as described below for large-
area tips. The AFM tips used for the bulk of norbornylog
resistance measurements were plateau-style tips purchased from
Pacific Nanotechnology. Tips and silicon substrates were coated
with 50 Å Cr and 1000 Å Au in a thermal evaporator at ∼1 ×
8
-9 Å offset between the measured value of film thickness for SAMs
of 4 and the expected thickness. Error bars represent one standard
deviation from the mean. Also shown are the thicknesses of four
individual SAM samples (one sample of each chain length) determined
by XPS (open circles) and ellipsometry (filled squares) for a direct
comparison of thickness between the two techniques. Good correlation
is shown between thickness obtained by XPS and thickness obtained
by ellipsometry.
length increases. To address these issues, we have extensively
characterized the thickness, molecular orientation, and surface
coverage of the films using spectroscopic ellipsometry, XPS,
RAIRS, and RBS.
-
6
1
0
Torr base pressure. After use, some Au-coated tips were
characterized by scanning electron microscopy (SEM) using a
JEOL 6500 field-emission gun SEM operating at 10 kV.
Current-voltage measurements were made using a modified
MultiMode AFM (Digital Instruments). Junctions were formed
by bringing Au-coated AFM tips into contact with SAMs on
Au substrates. A Keithley 236 source/measure unit was used
to apply a voltage to the AFM tip, and the substrate was
grounded through a Keithley 6517 DC ammeter. For a typical
data taking session with a single Au-coated tip, several I-V
traces per SAM were measured, giving an average resistance
value and standard deviation for each molecular chain.
Monolayer Thickness. Film thicknesses for the various chain
lengths determined by ellipsometry are shown in Figure 3. The
thicknesses of SAMs of 1-3 increase linearly with the number
of carbons in the chain. One would expect 4 to follow this trend
also, because it is a homologue of the previous molecules.
However, the thickness of SAMs of 4 is ∼8 Å less than
predicted by extending the trendline over an additional four
carbon atoms. Also shown in Figure 3 are film thicknesses
determined by XPS. The XPS results clearly corroborate the
ellispometry data, and show again that the thickness of SAMs
of 4 is less than the predicted thickness. The discrepancy in the
thickness of SAMs of 4 with respect to the measured thicknesses
of SAMs of 1-3 suggests two possibilities. The first is that the
inconsistency in measured thickness represents a decrease in
the average tilt angle of molecules in SAMs of 4 with respect
to 1-3. The second possibility is that 4 is long enough to hairpin
to the surface. To distinguish between a change in tilt angle
and a change to a hairpinned structure, we analyzed films by
RAIRS, RBS, and XPS.
Results and Discussion
Film quality is a critical factor in CP-AFM studies based on
SAMs. At the start of this work, it was not clear whether
molecules 1-4 would form ordered monolayers. A concern was
that the bicyclic nature of the bridges forces each chain to
possess a large effective cross-section, which could potentially
result in disorder because the molecules cannot pack as tightly
as in the alkanethiol system. Further, energy minimizations have
5
1
shown that these rigid molecules exhibit a curved backbone,
which should further decrease packing efficiency, as illustrated
in Figure 2. Finally, the R,ω-functionalization with thiol groups
could allow the molecules to bind to the surface at both ends
Monolayer Surface Coverage. Infrared spectra of organic
monolayers can be analyzed to obtain a relative packing density
and also to determine molecular orientation. We have compared
the relative peak areas of the primary methoxy C-O stretch in
bulk samples of 1-4 to their corresponding values in SAMs.
(hairpin); the probability of such an event increases as the chain

5210 J. Phys. Chem. B, Vol. 109, No. 11, 2005
Beebe et al.
Figure 5. (A) RBS spectrum of 2 on Au-coated silicon wafer. The
sulfur peak is easily resolved from both the baseline and the Si peak.
(
B) RBS spectra of molecules 1-4 showing the sulfur region. Baselines
Figure 4. (A) Bulk phase FTIR absorbance spectrum of the norbor-
have been adjusted to coincide and Gaussian fits were used to determine
peak area. The inset shows surface coverage as a function of molecular
chain length.
nylogous molecules in a KBr matrix. The inset shows linear scaling of
-
1
the C-O stretch at 1114 cm with the number of methoxy groups per
molecule. (B) RAIRS spectra of 2-4 show a nonlinear scaling of the
C-O stretch with the number of methoxy groups. The decreased area
of 4 supports a decrease in surface coverage.
Because there are no published procedures for measuring the
sulfur content of SAMs via RBS, it is important to discuss the
measurement subtleties involved. Initally, spectra were collected
with the incident beam in a random rotation configuration (there
is a small amplitude “jitter” on beam position). This configu-
ration is often used to minimize the effects of channeling through
the silicon substrate. However, spectra collected in this manner
exhibited a sulfur peak that was barely distinguishable from
the Si background signal. Changing to fixed position scans
resulted in an increase in signal-to-background ratio, making
that the preferred method for data collection. As described in
the Experimental Section, 10 acquisitions of 20 µC each were
added together to improve the signal. Care was also taken to
maximize the amount of channeling, which minimizes the
contribution of the Si peak in the measured spectrum, and results
in a further increase in S/N ratio. Total acquisition time per
film was approximately 5 h. A typical spectrum illustrating the
relative peak intensities of sulfur and the elements present in
the substrate is shown in Figure 5A. An expansion of the sulfur
region (Figure 5B) shows a decrease in peak area for SAMs of
In the case of the bulk-phase measurements, care was taken to
determine the exact amount of sample inserted into the KBr
matrix. This allows us to make molar absorptivity measurements
of the methoxy absorption. As shown in Figure 4A, the molar
absorptivity associated with the methoxy peak at 1114 cm- in
the bulk phase scales linearly with the number of methoxy
groups in the molecule. Because the molecules in question form
a homologous series, we would expect the same type of scaling
in the case of the monolayer spectra if both the surface coverage
and molecular orientation remain constant across the series.
However, as shown in Figure 4B, there is an obvious difference
in the spectrum of SAMs of 4 when compared to SAMs of 1-3
1
-
1
-
the primary methoxy peak at 1114 cm has been damped.
The area of this peak is now essentially equal to the peak area
associated with SAMs of 3. This decrease in peak area suggests
that there are less methoxy groups (and therefore fewer
molecules) on the surface in the case of 4 than there are for
SAMs of 1-3, i.e., that the surface coverage of 4 is lower than
that of 1-3.
4
in relation to SAMs of the other molecules. Peak area (A) is
directly related to surface coverage (NS) through the following
equation:
To determine the surface coverage quantitatively, we have
used RBS to measure the surface concentration of sulfur atoms.
There are only a few literature examples involving RBS analysis
A
σΩQ
52-54
N )
(2)
of organic monolayers,
and the focus of those papers is
S
not on determining sulfur content. To our knowledge, we present
the first experimental evidence that RBS is an effective
technique for measuring the surface coverage of thiol-based
SAMs.
where σ is the average scattering cross section, Ω is the detector
solid angle, and Q is the total number of collected He particles.
It is clear from Figure 5B that the signal-to-noise ratio is still
+

SAMs of Norbornane Homologues on Au
J. Phys. Chem. B, Vol. 109, No. 11, 2005 5211
determine whether the molecules are standing up or lying down.
Because there are four sulfur atoms per molecule, with two at
each end, the area of the bound peak should equal that of the
unbound peak if there is no hairpinning to the metal surface.
However, the areas that one must compare are not those initially
measured, because the electrons emanating from the bound
sulfur atoms are attenuated by the carbon backbone of the SAM
before reaching the detector. The area under the bound peak
must be normalized according to the following equation:
A0
A )
(3)
-
λ cos θ
z
exp
(
)
where A and A0 are the normalized and measured areas of the
bound sulfur peak, respectively, and the other variables are
defined as in eq 1. In calculating the percent of molecules doubly
bound to the surface, we have assumed that the oxidized sulfur
peak arises only from sulfur atoms that are not bound to the
metal surface. This may not be strictly true, as other groups
have shown that surface-bound sulfur atoms do oxidize.
However, it is likely that oxidation of the terminal sulfur atoms
is more facile than oxidation of the bound atoms, because for
oxygen to reach the bound atoms, it must diffuse through the
SAM matrix. This leads us to believe that the probability for
oxidation of the terminal sulfur atoms is much greater, and so
we add the oxidized peak area to the unbound peak area in a
calculation of the percentage of hairpinned molecules. Once the
normalized areas of bound and unbound peaks are known, the
percent of hairpinned molecules can be calculated thusly:
Figure 6. High-resolution XPS scans of the S 2p region in SAMs of
1
-4 show the presence of three different types of S atoms: bound
(∼161.8 eV), unbound (∼163.3 eV), and oxidized (∼167.9), in each
SAM. In each scan, the points represent the acquired data, the thin
black lines show the contribution to the overall spectrum from each of
the sulfur types, and the heavy black line shows the convolved fit
spectrum. Note that in SAMs of 1-3, the heights of the peaks
corresponding to bound S and unbound S are nearly equivalent; in
SAMs of 4, the bound S peak clearly dominates.
not ideal. For this reason, we fit Gaussians to the sulfur peaks
to quantitatively determine the surface coverage of sulfur by
determining the peak area of the Gaussian fit. The molecular
surface coverage was then calculated by dividing the sulfur
coverage by four, as there are four S atoms per molecule. The
results of this calculation are shown in the inset in Figure 5B.
The data show an essentially constant surface coverage for
2
molecules 1-3 (1.88-1.90 molecules/nm ), and a 6% decrease
2
areabound - (areaunbound + areaoxidized
)
for 4 (1.78 molecules/nm ), consistent with the results of
%
hairpinned )
(4)
ellipsometry and FTIR. Equation 2 does not explicitly account
for the roughness of the surface. However, we calculate a surface
total peak area
2
2
area of a 1.02 µm for a 1 µm image of a representative Au
surface, suggesting that the correction to the surface coverage
resulting from the surface roughness is very small.
The results of the above calculation are compiled in Table 1.
Because the noise in the XPS spectrum is not ideal, and because
we are making the assumption that the oxidized sulfur peak
arises totally from the terminal sulfur atoms, we estimate the
error in the percentage of hairpinned molecules to be (20%.
On the basis of the linear scaling of the methoxy peak area in
FTIR measurements, the linear thickness increase as monitored
by ellipsometry and XPS, and the similar surface coverage as
determined by RBS, it is likely that SAMs of 1-3 have a similar
percent of hairpinned molecules (i.e., the different percentages
of hairpinned molecules for SAMs of 1-3 in Table 1 are not
statistically significant). The small amount of hairpinning
predicted by XPS analysis coupled with the high degree of error
leads us to believe that 1-3 form SAMs that are bound to the
surface only on one end (i.e., 0% hairpinning). The conclusion
that 1-3 stand upright is further bolstered by a simple analysis
of the chain tilt angle using the ellipsometric data. Using
molecular lengths calculated by a B3LYP energy minimization
and the average SAM thickness from ellipsometry, we calculate
the tilt angles of 1-3 to be 39, 34, and 34°, respectively,
Film Orientation by XPS. The analyses by FTIR and RBS
indicate a smaller surface coverage of 4 with respect to 1-3,
but they do not clarify whether this decrease in surface coverage
corresponds to a higher degree of molecular tilt with respect to
the surface normal or to molecular hairpinning. To address this
issue, we have examined the sulfur binding energy (BE) by XPS.
XPS has been used extensively in thiol SAM characterization
to determine the chemical environment of the sulfur atoms (i.e.,
whether they are bound to the surface).5 High-resolution scans
of the S 2p region are shown in Figure 6. For each of the four
SAMs, sulfur exists in three different states: bound (BE ∼ 161.8
eV), unbound (BE ∼ 163.5 eV), and oxidized (BE ∼ 167.8
eV). As stated in the Experimental Section, we employed a peak
fitting procedure using Gaussian and Lorentzian line shapes to
deconvolute the measured spectra into contributions from the
bound and unbound sulfur states. By comparing the area under
the bound peak with that of the unbound peak, it is possible to
5,56
TABLE 1: Thicknesses and Sulfur Peak Areas for the Norbornylog SAMs Determined by XPS
sulfur peak areas
XPS
thickness (Å)
bound
(measured)
bound
(normalized)
percent
hairpinned
a
molecule
oxidized
unbound
1
2
3
4
6.2
9.5
12.2
10.4
722
633
512
349
3042
2601
2686
1109
3425
2149
2709
2398
4409
3162
4455
3663
8%
-1%
16%
43%
a
The error associated with this value is (20% (i.e., the percent of hairpinned molecules in SAMs of 4 ranges from 23 to 63%).

5212 J. Phys. Chem. B, Vol. 109, No. 11, 2005
Beebe et al.
Figure 9. (A) Schematic of the CP-AFM experiment performed with
plateau-style AFM tips illustrating the tip-molecule-substrate junction.
(B) SEM micrograph of the large-area plateau on the end of the AFM
tip.
Figure 7. (Top) two possible film orientations for molecules in
norbornylog SAMs. Molecules are either bound to the surface at one
end, as shown in the left half of the top panel, or they are bound twice
of variance associated with the trace-to-trace resistance measured
for a given tip. On the basis of prior experiments, we would
expect only about a factor of 5 lower current for the octane-
dithiol, which is within the variance of our measurement.
Therefore it is not surprising that the I-V traces of the two
molecules are essentially the same. The similarity in measured
resistance values suggests that the same transport mechanism
occurs through each of the two SAMs. Prior experiments on
SAMs with alkyl backbones have shown the dominant transport
mechanism to be nonresonant tunneling (assisted by the
(hairpinned), as shown on the right. It is unclear whether mixed-phase
SAMs in which some molecules stand up and some are hairpinned
exist as phase-separated monolayers (top panel) or as integrated
mixtures of standing up and lying down molecules (bottom panel).
3
molecular HOMO). Thus, the transport mechanism in nor-
bornylogs is also likely to be HOMO-mediated tunneling. The
similarity of measured currents between 1 and octanedithiol also
suggests that the terminal S atoms in norbornylog SAMs are
able to form chemical bonds with the AFM tip. This follows
because in prior experiments, we have observed 2 orders of
magnitude less current in SAMs of molecules terminated by a
methyl group than those terminated by a sulfur atom.29 The
CH3-tip contact is physical in nature, and results in poor orbital
overlap between atoms in the tip and atoms in the molecular
chain. The S-tip contact is a chemical bond, and results in greatly
increased molecule-electrode coupling and therefore decreased
contact resistance. If the terminal S atoms in 1 were unavailable
for bonding to the tip (i.e., the curved molecular backbone
caused the S atoms to become partially buried in the sea of
carbon atoms), we would expect I-V measurements on SAMs
of 1 to show greatly reduced current with respect to octane-
dithiol. Because that is not the case, it appears that SAMs of
1-4 form S-Au bonds on both sides of the molecular junc-
tion.
The second set of CP-AFM experiments we performed was
designed to test the length dependence of conductance in the
norbornylog series. In past experiments, we have typically used
metal-coated Si3N4 contact-mode tips that have a radius of ∼50
nm. However, the longest aliphatic chains we can reliably
measure using such tips are on the order of 12 carbon atoms
(e.g., dodecanethiol). Voltage sweeps on longer molecules result
in currents that are <1 pA, which is below our current
sensitivity. Two of the norbornylogous molecules (3 and 4) are
longer than 12 carbon atoms, which required the use of tips
with larger contact areas. We used plateau-style tips from Pacific
Nanotechnology that consist of an AFM tip that has been
truncated perpendicular to its z-axis, resulting in a structure that
has a micron-sized plateau used to contact the SAMs as shown
in Figure 9. It is not the case that the tip is atomically flat over
the entire plateau (i.e., there may be metallized surface particles
that make contact to the molecules); however, the resulting
contact area is hundreds of times larger than in the case of
contacts made with standard AFM tips. Because the number
and size of metallized asperities varies from tip to tip, the contact
area between tips can vary by orders of magnitude, thus causing
Figure 8. Single AFM tip used to acquire current-voltage data from
SAMs of 1 (filled circles) and SAMs of octanedithiol (open circles).
The similarity in measured current values between the two molecules
suggests that they have similar electronic structures. The fact that the
measured current for 1 is on par with that measured for octanedithiol
also suggests that the terminal S atoms of 1 make contact with the
AFM tip (see text).
assuming no hairpinning in each case. The similarity of the tilt
angles suggests that SAMs of 1-3 have similar orientation, and
no change in the percentage of hairpinned molecules from one
SAM to the next. A similar calculation for molecules of 4 gives
a tilt angle of 55°. It is obvious from every analysis technique
that SAMs of 4 are different from SAMs of 1-3, and we are
confident on the basis of the data from ellipsometry, FTIR, RBS,
and XPS that SAMs of 4 have some fraction of molecules
doubly bound to the substrate metal surface (Figure 7).
Electrical Characterization by CP-AFM. As stated in the
Experimental Section, two types of AFM experiments were
performed. In the first, we coated a standard pyramidal AFM
tip with Au for use as a nanoscale spring-loaded top contact.
We brought this electrode into contact with SAMs on Au under
2
nN applied load, and applied a voltage sweep to the tip while
the substrate was held at ground. The purpose of this set of
measurements was to verify that SAMs of 1-4 could indeed
be used for CP-AFM experiments, and to compare their
current-voltage characteristics with those of a system more
familiar to us: alkanedithiol SAMs. Figure 8 shows a com-
parison between the I-V behavior of 1 and octanedithiol on
Au. Both I-V curves are linear over a 0.5 V sweep range, and
the two molecules have very similar resistance values. Octane-
dithiol has one more C-C bond along its molecular axis than
does 1, which suggests that it should have a slightly higher
resistance that is not reflected in the data in Figure 8. However,
due to substrate roughness, there is at least an order of magnitude

SAMs of Norbornane Homologues on Au
J. Phys. Chem. B, Vol. 109, No. 11, 2005 5213
Figure 10. Current-voltage behavior of a CP-AFM junction incor-
porating 4. The dashed line is a Simmons fit with parameters: A )
-
3
2
2
.3 × 10 µm , L ) 17.5 Å, æ ) 1.8 eV.
the current to vary accordingly. For that reason, we have used
a single tip to electrically characterize SAMs of 1-4.
Figure 10 shows the sigmoidal dependence of current on
voltage for a SAM of 4 on Au. This trace is similar in shape to
I-V characteristics observed for SAMs with alkyl backbones.
The dashed line is a fit to the Simmons equation for tunneling
through a rectangular barrier:57
qA
qV
2
x
2m(φ - qV)
I )
φ -
exp -
L -
[
]
2
2
{
(
)
2
p
4
π pL
qV
2
x
2m(φ + qV)
φ +
exp -
L
(5)
[
]
}
(
)
2
p
where φ is the height of the barrier, L is the barrier width, m is
the electron mass, and A is the electrode area. The reasonable
fit provides qualitative support for nonresonant tunneling as the
transport mechanism, though it is well-known that the Simmons
model is only a crude approximation for nonresonant tunneling
in molecular junctions.29 Though an exact assignment of the
transport mechanism as nonresonant tunneling is impossible
without temperature dependent measurements, more evidence
for that mechanism is found in Figure 11. Figure 11A shows
that over small voltage sweeps, the traces are linear and the
slopes clearly decrease with increasing chain length. Figure 11B
shows that the current decreases exponentially with chain length
as expected for nonresonant tunneling.
Figure 11. (A) Measured currents for 1-4 are linear over the low-
voltage regime, and currents decrease with increasing chain length. (B)
Current scales exponentially with chain length, consistent with a
nonresonant tunneling mechanism.
The key result of this study is shown in Figure 12, which
shows a semilog plot of low voltage resistance versus molecular
contour length (number of carbons) for SAMs of 1-4. In the
low-voltage regime, the characteristic equation for nonresonant
tunneling is
Figure 12. Semilog plot of resistance versus molecular length for
junctions based on SAMs of 1-4. Although the SAM thickness does
not increase monotonically with chain length, the resistance scales
exponentially with the number of carbon atoms in the chain.
Note that it is perhaps surprising that the data in Figure 12
nicely fit eq 6. For example, one might have expected that the
resistance would scale exponentially with SAM thickness instead
of molecular contour length (i.e., a through-space transport
mechanism instead of through-bond). However, it is important
to keep in mind that the apparent thickness of SAMs of 4 is the
result of averaging over domains of upright and hairpinned
molecules. Assuming 43% of the molecules in SAMs of 4 are
hairpinned, and a 36° tilt angle for the nonhairpinned molecules,
we predict a thickness of 8 Å for hairpinned molecules and a
thickness of 19 Å for nonhairpinned molecules. Thus it is
possible that the tip only contacts the molecules that are singly
bound because they are more than twice as tall as the doubly
bound molecules. In this scenario, substitution of a junction
length of 19 Å, a contact resistance of 6000 Ω, and a â value
R ) R exp(âs)
(6)
0
where R is the junction resistance, R0 is the contact resistance,
â is the tunneling attenuation factor, and s is the junction length.
Resistance values plotted in Figure 12 were obtained from linear
fits to several low-voltage I-V curves of each molecule, which
generated a resistance and standard deviation for each chain
length. There is a clear exponential dependence of resistance
on chain length for all molecules. This is consistent with eq 6
for nonresonant tunneling. From the slope of the graph in Figure
-
1
-1
1
2 we extract a value of 1.1 C (0.9 Å ) for â. The â value
is in excellent agreement with â values determined for saturated
alkane chains in molecular junctions. The similarity of â values
between the norbornyl and alkyl systems suggests that the
electronic details of molecular junctions formed from SAMs
of each type of molecule (i.e., charge transfer at the metal-
molecule interface, relative position of the molecular energy
levels with respect to the system Fermi level, etc.) are also
similar.
-
1
of 0.9 Å into eq 6 gives a theoretical junction resistance of 2
1
1
× 10 Ω, exactly the measured resistance of the junction. This
suggests that it is definitely possible for the AFM tip to be
contacting only the molecules that stand up in mixed-phase

5214 J. Phys. Chem. B, Vol. 109, No. 11, 2005
Beebe et al.
SAMs, and therefore it is reasonable for the junction resistance
to scale with the contour length of the molecules in Figure 12.
Conversely, if the tip could make contact with hairpinned
molecules, the contact point would be somewhere in the middle
of the molecular backbone. This has two consequences: there
is likely now a tip-methylene contact instead of a tip-sulfur
contact, and the probable tunneling distance is half what it would
be for an upright molecule. Prior experiments in our laboratories
have shown a factor of 100 increase in resistance for metal-
CH3 contacts compared to metal-S contacts. However, that
small increase pales in comparison to the decreased resistance
that would arise from a drastic decrease in the length of the
tunneling junction. Using the thickness values calculated earlier,
and a contact resistance of 600 000 Ω, we calculate the junction
(5) Kushmerick, J. G.; Lazorcik, J.; Patterson, C. H.; Shashidhar, R.;
Seferos, D. S.; Bazan, G. C. Nano Lett. 2004, 4, 639-642.
(6) Pollack, S. K.; Naciri, J.; Mastrangelo, J.; Patterson, C. H.; Torres,
J.; Moore, M.; Shashidhar, R.; Kushmerick, J. G. Langmuir 2004, 20, 1838-
1842.
(7) Chabinyc, M. L.; Chen, X.; Holmlin, R. E.; Jacobs, H.; Skulason,
H.; Frisbie, C. D.; Mujica, V.; Ratner, M. A.; Rampi, M. A.; Whitesides,
G. M. J. Am. Chem. Soc. 2002, 124, 11730-11736.
(
8) Liu, Y.-J.; Yu, H.-Z. ChemPhysChem 2003, 4, 335-342.
(9) Solak, A. O.; Ranganathan, S.; Itoh, T.; McCreery, R. L. Electro-
chem. Solid-State Lett. 2002, 5, E43-E46.
(
10) Slowinski, K.; Slowinska, K. U.; Majda, M. J. Phys. Chem. B 1999,
03, 8544-8551.
11) Bumm, L. A.; Arnold, J. J.; Dunbar, T. D.; Allara, D. L.; Weiss,
P. S. J. Phys. Chem. B 1999, 103, 8122-8127.
12) Bumm, L. A.; Arnold, J. J.; Cygan, M. T.; Dunbar, T. D.; Burgin,
T. P.; Jones, L., II.; Allara, D. L.; Tour, J. M.; Weiss, P. S. Science 1996,
1
(
(
2
71, 1705-07.
9
resistance to be 1 × 10 Ω for the scenario where the tip contacts
(13) Ishida, T.; Mizutani, W.; Choi, N.; Akiba, U.; Fujihira, M.;
both hairpinned and nonhairpinned molecules.5 This is 2 orders
of magnitude less than the measured resistance, and thus it is
unlikely that the tip is contacting both hairpinned and nonhair-
pinned molecules simultaneously.
8
Tokumoto, H. J. Phys. Chem. B 2000, 104, 11680-11688.
14) Guisinger, N. P.; Greene, M. E.; Basu, R.; Baluch, A. S.; Hersam,
M. C. Nano Lett. 2004, 4, 55-59.
15) Datta, S.; Tian, W.; Hong, S.; Reifenberger, R.; Henderson, J. I.;
Kubiak, C. P. Phys. ReV. Lett. 1997, 79, 2530-2533.
16) Gomar-Nadal, E.; Ramachandran, G. K.; Chen, F.; Burgin, T.;
Rovira, C.; Amabilino, D. B.; Lindsay, S. M. J. Phys. Chem. B 2004, 108,
213-7218.
17) Ishida, T.; Mizutani, W.; Liang, T.-T.; Azehara, H.; Miyake, K.;
Sasaki, S.; Tokumoto, H. Ann. N. Y. Acad. Sci. 2003, 1006, 164-186.
18) Wold, D. J.; Haag, R.; Rampi, M. A.; Frisbie, C. D. J. Phys. Chem.
(
(
(
Conclusions
7
(
We have synthesized a series of thiol-terminated norbornylogs
and formed SAMs of them on Au from CH2Cl2. SAMs were
characterized structurally using ellipsometry, FTIR, RBS, and
XPS, and electrically with CP-AFM. Structural characterization
shows that 1-3 form similarly oriented monolayers, bound to
the Au surface only on one end, with an average tilt angle of
(
B 2002, 106, 2813-2816.
(19) Cui, X. D.; Primak, A.; Zarate, X.; Tomfohr, J.; Sankey, O. F.;
Moore, A. L.; Moore, T. A.; Gust, D.; Nagahara, L. A.; Lindsay, S. M. J.
Phys. Chem. B 2002, 106, 8609-8614.
(
20) Cui, X. D.; Primak, A.; Zarate, X.; Tomfohr, J.; Sankey, O. F.;
36° with respect to the surface normal. Molecules of 4 form
Moore, A. L.; Moore, T. A.; Gust, D.; Harris, G.; Lindsay, S. M. Science
2001, 294, 571-574.
SAMs with a structure different from those of 1-3, in which a
fair percentage of molecules are bound to the Au substrate at
both ends. By using large-area AFM tips to make I-V
measurements, we have probed conductance through norborny-
logs 1-4, which range from 13 to 28 Å in length. Electrical
characterization shows strong evidence for a charge transport
mechanism of nonresonant tunneling, as determined by the
exponential dependence of resistance on molecular length.
Although SAMs of 4 contain some hairpinned molecules, our
results show that the resistance of junctions of SAMs of 4 scales
as if there were no hairpinned molecules, likely because the tip
does not make contact with molecules that are not standing up.
(
21) Le, J. D.; He, Y.; Hoye, T. R.; Mead, C. C.; Kiehl, R. A. Appl.
Phys. Lett. 2003, 83, 5518-5520.
(
22) Chen, J.; Wang, W.; Reed, M. A.; Rawlett, A. M.; Price, D. W.;
Tour, J. M. Appl. Phys. Lett. 2000, 77, 1224-1226.
(23) Metzger, R. M. Chem. ReV. 2003, 103, 3803-3834.
(
(
(
24) Xu, B.; Zhang, P.; Li, X.; Tao, N. Nano Lett. 2004, 4, 1105-1108.
25) Xiao, X.; Xu, B.; Tao, N. J. Nano Lett. 2004, 4, 267-271.
26) Luo, Y.; Collier, C. P.; Jeppesen, J. O.; Nielsen, K. A.; Delonno,
E.; Ho, G.; Perkins, J.; Tseng, H.-R.; Yamamoto, T.; Stoddart, J. F.; Heath,
J. R. ChemPhysChem 2002, 3, 519-525.
(
27) Lau, C. N.; Stewart, D. R.; Williams, R. S.; Bockrath, M. Nano
Lett. 2004, 4, 569-572.
28) Wold, D. J.; Frisbie, C. D. J. Am. Chem. Soc. 2000, 122, 2970-
2971.
(
(
29) Engelkes, V. B.; Beebe, J. M.; Frisbie, C. D. J. Am. Chem. Soc.
2
004, 126, 14287-14296.
Acknowledgment. J.M.B. thanks John C. Nelson of the
University of Minnesota Characterization Facility and Yong-
qiang Wang now with Los Alamos National Labs for assistance
with spectroscopic ellipsometry and RBS measurements, re-
spectively. C.D.F. thanks the National Science Foundation
(30) Beebe, J. M.; Engelkes, V. B.; Miller, L. L.; Frisbie, C. D. J. Am.
Chem. Soc. 2002, 124, 11268-11269.
(31) Tong, G. S. M.; Kurnikov, I. V.; Beratan, D. N. J. Phys. Chem. B
2002, 106, 2381-2392.
(32) Winkler, J. R.; Di Bilio, A. J.; Farrow, N. A.; Richards, J. H.; Gray,
H. B. Pure Appl. Chem. 1999, 71, 1753-1764.
(DMR-0084404, CHE-0315165) for financial support. The
(33) Goldman, C. Phys. ReV. A: Ato., Mol., Opt. Phys. 1991, 43, 4500-
Australian group thanks the Australian Research Council for
support, and the Australian Partnership for Advanced and the
Computing (APAC) and the Australian Centre for Advanced
Computing and Communications (AC3) for allocation of
computing time.
9.
(34) Jones, M. L.; Kurnikov, I. V.; Beratan, D. N. J. Phys. Chem. A
2002, 106, 2002-2006.
(
35) Kelley, S. O.; Barton, J. K. Science 1999, 283, 375-381.
(36) Arkin, M. R.; Stemp, E. D. A.; Holmlin, R. E.; Barton, J. K.;
Hormann, A.; Olson, E. J. C.; Barbara, P. F. Science 1996, 273, 475-
80.
37) Seischab, M.; Lodenkemper, T.; Stockmann, A.; Schneider, S.;
4
(
Supporting Information Available: Experimental details
for the synthesis of compounds 1-4. This information in
available free of charge at http://pubs.acs.org.
Koeberg, M.; Roest, M. R.; Verhoeven, J. W.; Lawson, J. M.; Paddon-
Row, M. N. Phys. Chem. Chem. Phys. 2000, 2, 1889-1897.
(38) Paddon-Row, M. N.; Verhoeven, J. W. New J. Chem. 1991, 15,
107-16.
(39) Oevering, H.; Paddon-Row, M. N.; Heppener, M.; Oliver, A. M.;
References and Notes
Cotsaris, E.; Verhoeven, J. W.; Hush, N. S. J. Am. Chem. Soc. 1987, 109,
3258-69.
(40) Balaji, V.; Ng, L.; Jordan, K. D.; Paddon-Row, M. N.; Patney, H.
K. J. Am. Chem. Soc. 1987, 109, 6957-69.
(41) Warman, J. M.; De Haas, M. P.; Paddon-Row, M. N.; Cotsaris,
E.; Hush, N. S.; Oevering, H.; Verhoeven, J. W. Nature 1986, 320, 615-
16.
(
1) Reichert, J.; Weber, H. B.; Mayor, M.; von Lohneysen, H. Appl.
Phys. Lett. 2003, 82, 4137-4139.
2) Kergueris, C.; Bourgoin, J. P.; Palacin, S.; Esteve, D.; Urbina, C.;
Magoga, M.; Joachim, C. Phys. ReV. B 1999, 59, 12505-12513.
(
(
3) Wang, W.; Lee, T.; Reed, M. A. Phys. ReV. B 2003, 68, 035416/
1
-035416/7.
(4) Wang, W.; Lee, T.; Kretzschmar, I.; Reed, M. A. Nano Lett. 2004,
(42) Hush, N. S.; Paddon-Row, M. N.; Cotsaris, E.; Oevering, H.;
4
, 643-646.
Verhoeven, J. W.; Heppener, M. Chem. Phys. Lett. 1985, 117, 8-11.

SAMs of Norbornane Homologues on Au
J. Phys. Chem. B, Vol. 109, No. 11, 2005 5215
(
43) Wold, D. J.; Frisbie, C. D. J. Am. Chem. Soc. 2001, 123, 5549-
556.
44) Raymo, F. M.; Alvarado, R. J.; Pacsial, E. J.; Alexander, D. J.
Phys. Chem. B 2004, 108, 8622-8625.
45) Weckenmann, U.; Mittler, S.; Naumann, K.; Fischer, R. A.
Langmuir 2002, 18, 5479-5486.
46) Brower, T. L.; Garno, J. C.; Ulman, A.; Liu, G.-y.; Yan, C.;
(52) Wohlfart, P.; Weiss, J.; Kashammer, J.; Winter, C.; Scheumann,
V.; Fischer, R. A.; Mittler-Nehera, S. Thin Solid Films 1999, 340, 274-
279.
5
(
(53) Winter, C.; Kashammer, J.; Mittle-Neher, S.; Fischer, R. A. Opt.
Mater. (Amsterdam) 1998, 9, 352-355.
(
(54) Colvin, V. L.; Goldstein, A. N.; Alivisatos, A. P. J. Am. Chem.
Soc. 1992, 114, 5221-30.
(
Goelzhaeuser, A.; Grunze, M. Langmuir 2002, 18, 6207-6216.
(55) Deng, W.; Yang, L.; Fujita, D.; Nejoh, H.; Bai, C. Appl. Phys.
A-Mater. 2000, 71, 639-642.
(
47) Joo, S. W.; Han, S. W.; Kim, K. Langmuir 2000, 16, 5391-5396.
(48) Kohli, P.; Taylor, K. K.; Harris, J. J.; Blanchard, G. J. J. Am. Chem.
(56) Rieley, H.; Kendall, G. K.; Zemicael, F. W.; Smith, T. L.; Yang,
S. Langmuir 1998, 14, 5147-5153.
Soc. 1998, 120, 11962-11968.
49) Laibinis, P. E.; Bain, C. D.; Whitesides, G. M. J. Phys. Chem. 1991,
(
(57) Simmons, J. G. J. Appl. Phys. 1963, 34, 1793-1803.
(58) The contact resistance associated with the tip-norbornylog-
substrate junction was 6000 Ω. Under the assumption that contacting the
backbone of hairpinned molecules is similar to making contact with a CH3
group, we expect a contact resistance approximately 100 times greater than
in the case of the norbornylogs.
9
5, 7017-21.
(
50) Bain, C. D.; Whitesides, G. M. J. Phys. Chem. 1989, 93, 1670-3.
(51) Unconstrained geometry optimizations were carried out at the
B3LYP/6-31G(d) level of theory, using Gaussian 98, Revision A.7;
Gaussian, Inc: Pittsburgh, PA.